Question 1

What is ThinkingRoot?

Accepted Answer

ThinkingRoot is growing AI infrastructure that gives every user their own mind — it stores facts as a knowledge graph, returns grounded answers with citations, and lets you branch, merge, and audit memory like code.

Question 2

How is ThinkingRoot different from a vector database?

Accepted Answer

A vector database stores and retrieves chunks. ThinkingRoot also learns from which memories get used, predicts what happens next, consolidates memory while idle (dream), and forks and merges like git — with provenance on every fact and verifiable deletion.

Question 3

Is ThinkingRoot open source?

Accepted Answer

Yes. The ThinkingRoot engine is open source and self-hostable — run it in our cloud, your cloud, or your laptop. The managed cloud adds accounts, a secure gateway, and project provisioning.

Question 4

How does ThinkingRoot avoid hallucinations?

Accepted Answer

It answers only from what it can cite. If no stored claim verifies the answer, it says it does not know instead of guessing. We call this 'verified or silent.'

Question 5

What languages and tools does ThinkingRoot support?

Accepted Answer

A TypeScript and JavaScript SDK that runs in Node, Bun, Deno, edge, and the browser; a REST API; and a one-line MCP server that plugs into Claude, Cursor, and Codex.

Question 6

Can I delete what an AI knows about a user?

Accepted Answer

Yes. Because ThinkingRoot stores facts in a graph, you can remove a specific fact, and every cached answer that depended on it is evicted too — something impossible to do cleanly with knowledge baked into model weights.

ThinkingRoot Blog

CompAG: Compile-Augmented Generation

Introducing Flows in ThinkingRoot CompAG

Connected Memory: OAuth & API Integrations

Building a Zero-Leak Accuracy Benchmark